No one is disputing that Professor or your suggestion of a methodology.
What is a bit dubious is an advocacy of a particular outcome of your methodology.
He could of course have arranged for someone to drop a wafer into a pocket unannounced.
Of course we don't know the nature or methodology of his experiment.
I think the point being, firstly that (as I mentioned before) the vicar is probably the person most bought into the significance of the wafers as a priest, compared to a bunch of inmates. Secondly that he (rather than the inmates) is the only person reporting a change in behaviour, so it could be purely that to be his subjective perception, rather than an objective observation.
And in science we develop hypotheses, which are exactly what you complain of - in other words an expected outcome, that is tested and proven to stand up to the testing or rejected.
So it is perfectly reasonable to have a hypothesis that claims that 'carrying communion wafers alters the manner in which the vicar interacts with inmates' with secondary hypotheses that:
'the inmates alter their behaviour too' or alternatively 'that the vicar perceives an alteration in their behaviour due to his altered psychological state.
How would you test - well it would need to be a double blinded independently observed study. So you could have an envelope that may or may not contain wafers - whether it does or not is not known by the vicar, the inmates or the observer. You would then need an independent person to observe the interactions (obviously not the vicar nor the inmates). What would be interesting too would be, following the independent observations, to have another researcher interview (or use a questionnaire etc) to determine whether the inmates perceived a difference in the response of the vicar to them or a response of themselves to the vicar. And to do the same for the vicar.
In order to assess psychological effects you would repeat the experiment under circumstances where you tell the vicar, the inmates or both whether or not the wafers are in the envelope - and you would do this in a random manner, so sometimes what you tell them is true sometime not.
That would be how you would go about the experiment.